Luận án: Điều khiển & Tối ưu hóa Phân tán cho Microgrid với Hardware-in-the-Loop

THÈSE Pour obtenir le grade de DOCTEUR DE LA COMMUNAUTE UNIVERSITE GRENOBLE ALPES Spécialité : GENIE ELECTRIQUE Arrêté ministériel : 25 mai 2016 Présentée par Tung Lam NGUYEN Thèse dirigée par Raphael CAIRE, Maître de Conférences, GINP, G2ELAB et codirigée par Quoc Tuan TRAN, Professeur, CEA-INES préparée au sein du Laboratoire Laboratoire de Génie Electrique dans l'École Doctorale Electronique, Electrotechnique, Automatique, Traitement du Signal (EEATS) Contrôle et optimisation distribués basés sur l'agent dans les micro-réseaux avec implémentation Hardware-in-the-Loop Agent-based distributed control and optimization in microgrids with Hardware-in-the-Loop implementation Thèse soutenue publiquement le 22 mai 2019, devant le jury composé de : Madame Salima HASSAS Professeur, Department of Computer Science of Polytech Lyon, Claude Bernard-Lyon1 University, Président Monsieur Mohamed ELLEUCH Professeur, Ecole Nationale d'Ingénieurs de Tunis - Université de Tunis El Manar, Rapporteur Monsieur Geert DECONINCK Professeur, ESAT - ELECTA, Electrical Energy and Computer Architectures, KU LEUVEN, Rapporteur Monsieur Raphael CAIRE Maître de Conférences, Univ.Grenoble Alpes, G2Elab, Directeur de thèse Monsieur Quoc Tuan TRAN Professeur, CEA-INES, Co-directeur de thèse i This thesis is dedicated to my parents, my wife and my beloved daughter. Contents Abstract 1 Resume 3 I Introduction 5 I.1 Overview of Microgrids .a Concepts of Microgrids .b Structure of Microgrids .c Operation Modes of MGs .c-i Grid-connected Mode .c-ii Islanded Mode .d Advantages of MGs .2 Distributed Energy Resources .b Energy Storage Systems .3 Control of Power Converter in Microgrids .a General Conguration of Power Converter .b Classication of Power Converter in Microgrid .b-i Grid-forming Unit .b-ii Grid-feeding Unit .b-iii Grid-supporting Unit .c Operation Scenarios of Power Converter .4 Hierarchical Control Structure in Microgrids .5 Centralized vs Distributed control. 29 II The layer structure and the laboratory platform for dis- tributed control and optimization in microgrids 33 II.2 The Multi-agent system. 35 iv CONTENTS II.b The Multi-agent system in Microgrids .c MAS platform in power system domain .3 The layer structure .4 Laboratory Platform for the Validation of Distributed Algo- rithms .a Overview of the Real-time Simulation for Power Systems .a-i Digital Real Time Simulators .a-ii The Laboratory Setup for DRTS .b The Laboratory Platform for Distributed Control Validation .b-i Digital real-time simulator .b-ii CHIL with MAS platform and a realistic communication network.

47 III Agent-based distributed secondary control in islanded micro- grids 49 III.b Consensus algorithm in a graph system .3 Agent-based distributed control using finite-time consensus algorithm .a-i SoC Estimation of ESSs .a-ii Primary Control .a-iii Formulation of Secondary Control .b Distributed nite-time secondary control .b-i Finite-time SOC Balance Control .b-ii Frequency Restoration .b-iii Voltage Restoration .c Design the Agent for Distributed Finite-time Consensus .d Controller Hardware-in-the-loop Validation .d-i Scenario 1: Validation under Step Response .d-ii Scenario 2: Plug-and-Play Capability of the Agent System 68 III.d-iii Scenario 3: Comparison with Linear Control Method .4 Agent-based distributed control using average consensus al- gorithm .a Average Consensus Algorithm .b Design of Agent with the Plug and Play Feature .c-i Platform Design for Validation of Distributed Control in MG 78 III.c-ii Testing Procedure .c-iii Experimental Results. 92 IV Agent-based distributed optimal power flow in microgrids 95 IV.2 Alternating Direction Method of Multipliers .a General Structures for Distributed Optimization Problems .b Alternating Direction Method of Multipliers for General Distributed Prob- lems .3 Formulation of optimal power flow problem .a General Optimal Power Flow Formulation .b Distributed Optimal Power Flow Formulation .4 Distributed Optimal Power Flow using ADMM .5 Agent-based Distributed Optimal Power Flow in Grid-connected Microgrids Using ADMM .a Design of the Agent for Implementing ADMM .b-ii Controller Hardware-in-the-loop and Experimental Results 120 IV.6 Agent-based distributed optimal power flow in islanded micro- grids using ADMM .a Distributed Agent-Based Secondary and Tertiary Control Framework .a-i The Secondary Control .a-ii The Tertiary Control .c Controller Hardware-in-the-loop and Experimental Results. 144 Bibliographie 158 Publications 159 A Graph Theory 1 A.1 Connectivity of a Graph .2 Algebraic Graph Properties. 2 vi List of Figures I.1 The structure of MGs.2 Benet of MGs [1].3 Power converter for interfacing an power source to microgrid.4 Grid forming converter.5 Basic structure of controller of grid forming converter.6 Grid feeding converter.7 Basic structure of controller of grid feeding converter.8 Grid supporting converter.9 Basic structure of controller of grid supporting converter.10 The hierarchical control structure of MGs.11 Primary control strategies in microgrids.12 Simplied power ow diagram between two DG source and the grid.14 Frequency and voltage droop.15 Block diagram of the virtual output impedance method.16 Centralized control strategy in microgrids.17 Distributed control strategy in microgrids.18 Scheme of the thesis.1 Diagram of a general agent.2 The 2-layer structure.3 The proposed layer structure.4 Basic structure of HIL real time simulation [2, 3].5 Laboratory setup for DRTS.6 The distributed laboratory platform.8 Interface between RPI cluster and OPAL-RT.1 The two approaches of the implementation of the consensus algorithm for the distributed secondary control.2 Max-consensus in a network.3 Diagram of distributed nite-time consensus control framework for an ESS in the layer structure.4 Structure of the designed agent.5 The iterative process of the agent.7 Test case setup.

64 LIST OF FIGURES vii III.u unit; and (d) Frequency values measured at output of ESSs.u unit; and (d) Frequency values measured at output of ESSs.u unit; and (d) Frequency values measured at output of ESSs.11An undirected graph with 4 nodes and 4 edges .12Convergence of average consensus algorithm.13Diagram of distributed nite-time consensus for an ESS in the layer structure.14Diagram of distributed nite-time consensus for an ESS in the layer structure.15Diagram of distributed nite-time consensus for an ESS in the layer structure.16Algorithm of agent system when an ESS is out of MG.17Algorithm of agent system when an ESS is added to MG.18The MG test case study in the layer structure.19The designed laboratory platform.20Topology of MAS in each scenario. (a) The state values of agents; (b) Consensus values sent to controllers from agents; (c) Frequency values measured at output of ESSs; and (d) Active power output of ESSs. (a) The state values of agents; (b) Consensus values sent to controllers from agents; (c) Frequency values measured at output of ESSs; and (d) Active power output of ESSs.23Time of one loop consensus process Scenarios 12. (a) The state values of agents; (b) Consensus values sent to controllers from agents; (c) Frequency values measured at output of ESSs; and (d) Active power output of ESSs.

(a) The state values of agents; (b) Consensus values sent to controllers from agents; (c) Frequency values measured at output of ESSs; and (d) Active power output of ESSs. (a) The state values of agents; (b) Consensus values sent to controllers from agents; (c) Frequency values measured at output of ESSs; and (d) Active power output of ESSs.27Time of one loop consensus process Scenarios 3-5. (a) The state values of agents; (b) Consensus values sent to controllers from agents; (c) Frequency values measured at output of ESSs; and (d) Active power output of ESSs.29Time of one loop consensus process Scenario 6. (a) The state values of agents; (b) Consensus values sent to controllers from agents; (c) Frequency values measured at output of ESSs; and (d) Active power output of ESSs.

(a) The state values of agents; (b) Consensus values sent to controllers from agents; (c) Frequency values measured at output of ESSs; and (d) Active power output of ESSs. 92 viii LIST OF FIGURES IV.1 The relation of variables in a distributed problem. Coupling is represented by the global variable z, together with the constraints xk = Ek z, k = 1,. , K , where Ek projects global variables to the corresponding local variables.2 The graph of the example grid .3 The relation of global variables and local variables in agents in the example network.4 Diagram of an ESS for the distributed OPF in the layer structure.5 The structure of an agent implementing ADMM method for distributed OPF problem.6 The iterative process in a gen-agent and a load-agent.7 The test case grid-connected microgrid.8 The convergences of local variables and global variable in agents.9 The convergence of generator power on dierent values of ρ in agent 1.10The convergence of generator power on dierent values of ρ in agent 2.11The convergence of generator power on dierent values of ρ in agent 6.12The load active and reactive power .13The active power of ESS 2 calculated in agent 2 and measured from the simulation.14The reactive power of ESS 2 calculated in agent 2 and measured from the simulation.15The active power of ESS 6 calculated in agent 6 and measured from the simulation.16The reactive power of ESS 6 calculated in agent 6 and measured from the simulation.17The total active power losses.18The bus voltages.19The active and reactive power at the PCC point calculated in agent 1 and measured from the simulation.20Diagram of distributed secondary-tertiary framework for an ESS in the layer structure.21The agent structure for the distributed secondary-tertiary framework.22The iterative progresses of the two processes in a gen-agent.23The test case MG in the layer structure.24The convergence of active power calculation in agent 1 and the measurements at bus 1.25The convergence of active power calculation in agent 2 and the measurements at bus 2.26The convergence of active power calculation in agent 6 and the measurements at bus 6.27The active power outputs of the ESSs.28The voltages at all buses.29The total active power losses of the network.30The reactive power outputs of the ESSs from agent calculation and from simulation measurements.

141 LIST OF FIGURES ix A.1 Undirected and directed graph.2 Connected, disconnected and complete graph. 2 x List of Tables I.1 Types of distributed energy resources.2 Operational mode of power converter in dierent scenarios.1 Summary of features of common real time simulators for power systems [2, 3].1 Parameters of the test case microgrid.2 Parameters of the proposed control.3 Parameters of ESS inverter controllers.1 Parameters of 6-bus microgrid test case.2 The error metrics in dierent values of ρ and dierent iterations .3 The time for an ADMM loop in the grid-connected case.4 The time for an ADMM loop in the islanded case. 134 LIST OF TABLES xi Abstract In terms of the control hierarchy of a microgrid with multiple voltage source inverters, the coordination of local controllers is mandatory in the secondary and tertiary levels. In- stead of using a central unit as conventional approaches, in this work, distributed schemes are considered.

The distributed approaches have been taken attention widely recently due to the advantages of reliability, scalability, and security. The multi-agent system is an advanced technique having properties that make them suitable for acting as a basis for building modern distributed control systems. The thesis focuses on the design of agents for distributed control and optimization algorithms in microgrids with realistic on-line deploy- ment on a Hardware-in-the-Loop platform. Firstly, a three-layer structure is introduced to emphasize the operation of the agents and describe the relationships of agents and the physical system.

This concept divides a microgrid into three layers: the Agent layer, the Control layer, and the Device layer. Based on the provided architecture, a laboratory plat- form with Hardware-in-the-loop setup is constructed in the system level and can be used to test various distributed algorithms. This platform includes two parts: (1) a digital real- time simulator, which covers the Control layer and the Device layer, used to simulate test case microgrids with local controllers in real-time; and (2) a cluster of hardware Raspberry PIs is represented the multi-agent system operating in a sparse physical communication network. An agent is a Python-based program run on a single Raspberry PI owing abilities to transfer data with neighbors and computing algorithms to control the microgrid in a distributed manner.

The idea of the layer structure and the distributed laboratory plat- form are spread throughout the thesis for validating proposed algorithms and obtaining experimental results. Secondly, the distributed secondary controls in an islanded microgrid with two approaches are presented. In the rst approach of using a nite-time consensus algorithm, the microgrid is controlled with the improved performance to achieve multi- ple objectives of frequency/voltage restoration, the accuracy of proportional active power sharing and state of charge balance among energy storage systems.